I wonder if it is only about slew rate. The Stochino amp is VFb and achieves positive and negative slew rates in excess of 300v/uS. CFAs published on this forum have not yet come to such high slew rates.
Slew rate does not predict the imaging characteristic of an amplifier. What about Harmonic distortion and its spectral distribution?
Highly resolving amps do not make for the same subjective experiences; some are clinical and analytical whole others are simply more pleasant and musical without taking away resolution. Some amplifiers which mask inner details (read lacking in ultimate resolution) also tend to sound more musical.
So I am not sure if the CFA vs VFA debate is going to yield any acceptable results. In other words, the debate cannot be rendered simplistic.
At the end of the day, we need to come up with measurement factors which closely relate to a more accurate psycho-acoustic model than to rely upon the well known measurements for audio amplifiers.
Slew rate does not predict the imaging characteristic of an amplifier. What about Harmonic distortion and its spectral distribution?
Highly resolving amps do not make for the same subjective experiences; some are clinical and analytical whole others are simply more pleasant and musical without taking away resolution. Some amplifiers which mask inner details (read lacking in ultimate resolution) also tend to sound more musical.
So I am not sure if the CFA vs VFA debate is going to yield any acceptable results. In other words, the debate cannot be rendered simplistic.
At the end of the day, we need to come up with measurement factors which closely relate to a more accurate psycho-acoustic model than to rely upon the well known measurements for audio amplifiers.
Bob Cordell did it in a remarkably concise and detailed way in 1980 :
CordellAudio.com - Another View of TIM
It contains all the basic fundamentals of amplifiers behaviour towards frequency. Thirty three years later, they have not changed and should always be in memory of anyone involved in amplifier design.
The argued delay in feedback is dealt with.
Negative Feedback treats any unwanted signal at the output of an amplifier, be it from emanating from an internal cause, non linearity, power supply injection, or from an external cause, microphonic currents from moving voice-coils, induced hum in the same manner : NFB reduced it at each frequency by the amount of NFB at that frequency.
The demonstration that there is a deleterious delay in the process has never been done.
Concerning lab measurements, they are usually more severe than what amplifiers encounter when loaded by loudspeakers delivering music.
The test signals can me be made fast, non constant and non repetitive at will, in a word, made more demanding to amplifiers than any musical signal.
However, musical signals themselves can be used in a test which consists of extraction of distorsion : it consists of the subtraction of the input from the adequately attenuated and filtered output.
It was proposed by Baxandall, Hafler and in this Mike Renardson's thesis :
Distortion Measurement
It is difficult to imagine an objective test being closer to the reality.
This test is rarely used because an amp succeeding in conventional tests (let's says harmonic distorsion at 20 Hz and 20 kHz, and stable on 10 kHz square waves on reactive loads) and failing to the subtractive method has yet to be found.
There is never an instant current source. Every supply has an small inductive component, and if there is a fast current demand, the result is that there's a voltage drop across that inductance and what you see then is a ripple voltage on the supply line. The higher the frequency/faster the current draw, the larger the ripple. But the output signal will still be fine, unless you ask for so much output level that the amp clips 'into' the ripple.
Question: what sort of bandlimitimng do you use for your test square wave? 2 x audio frequency?
jan
I don't think anyone said instant current sources exist. However, the ripple you describe could also be coined voltage sag. And this is an effect you want to avoid applying to all your current loops. You just make the feedback work harder to counter for this effect and not to mention, at HF have this voltage modulation appear superimposed on the output too.
I remember you are the guy that said there's no such thing as an assumed infinite voltage with an ideal current source. While it's been mathematically been proven by mr. Norton and Kirchhoff.
Vertel me maar eens, wat is de open-klem spanning van een ideale stroombron?
Last edited:
Hi Jan !
Yes that`s truth ! ,
----------------------------------------------------------------------
but for example if I want to improve linear PSU DC energy storage capacity and in the same time DC energy delivering transient capacity than I connect many small values Elko`s in parallel instead of one big capacity Elko can , this usually improve Amp bass transient performance , but to improve transient response on highs than I connect 1-2 uF MKP bloc condenser from each output power transistor collector direct to main central star ground point , and yes there is no typical SS Amp PCB around at all for me , I made my DIY SS stuff same as for some tube Amps , using only Cu. solid core hard wired point to point wiring technique .
Square wave test I done with 20Khz max , but that`s for me just one electric test and nothing else , for me serious long term listening test with music program via good loudspeaker is much more important to observe real Amp performance , and is in the same time my final test .
Best Regards !
Last edited:
Still not following, if we design such that the excursions are limited to small deviations from the linear operating point where both curves are nearly the same what difference is there?
Thanks
-Antonio
You cant learn this way -- just not reading what has been put up for you to read. How are you going to understand CMA special set of conditions this way. Pls read what has already been done and recommended.
The threshold I used was -100dB. That was done with the topology shown and with bipolars and without SIM PC and software. Its ALL measured results. Then as is now. And it did not use a lot of gnfb!
Using similar topology in the 1980's the Hafler 101 preamp was spec'ed to be under .001% also. Its Done and even better understood now than then. try to catch up. This is about learning new tricks and topologies ... generically called The Current-Mode Perspective.
Today, I expect a lot better than what was done decades ago. But first undertand what was done - thru reading and thru SIm and then tests will go a long way to increasing the palet of design choices.
THx_RNMarsh
The threshold I used was -100dB. That was done with the topology shown and with bipolars and without SIM PC and software. Its ALL measured results. Then as is now. And it did not use a lot of gnfb!
Using similar topology in the 1980's the Hafler 101 preamp was spec'ed to be under .001% also. Its Done and even better understood now than then. try to catch up. This is about learning new tricks and topologies ... generically called The Current-Mode Perspective.
Today, I expect a lot better than what was done decades ago. But first undertand what was done - thru reading and thru SIm and then tests will go a long way to increasing the palet of design choices.
THx_RNMarsh
Last edited:
you are using as examples Class A output amps, with "small signal" Q of similar speed throughout, specifically for the output Q - basically "no-brainer" conditions for low distortion amplifier design
CFA preforming well is such examples says little about what I thought the debate was - does CFA topology have large, significant advantage in audio Power amps - with much slower, Class AB output stages limiting "speed" and distortion
I've often urged use of modern DSL driver CFA op amps at audio line level, or to drive dynamic headphones - but their internal Q are all from the same process, have similar (low GHz!) speed - and use a lot more Q - some cascode the input as well as the mirror - and early generation DSL driver op amps use mA output stage bias
CFA preforming well is such examples says little about what I thought the debate was - does CFA topology have large, significant advantage in audio Power amps - with much slower, Class AB output stages limiting "speed" and distortion
I've often urged use of modern DSL driver CFA op amps at audio line level, or to drive dynamic headphones - but their internal Q are all from the same process, have similar (low GHz!) speed - and use a lot more Q - some cascode the input as well as the mirror - and early generation DSL driver op amps use mA output stage bias
Last edited:
jcx,
Your statement about what you think this debate is about says it all. The debate was left in the other thread when we came to this new thread. This is not a debate thread, it is a thread to learn and understand the cfa circuit and how to build the best cfa that is possible, Leave the debate behind and either try and learn something, unless of course you don't need to and then why are you here? This isn't the bash the idea thread, please leave it at the door. If you have something useful to add to the cfa circuit great. It seems that you are impatient to get right to the final designs, but some of us are not anywhere near that level of understanding yet and I appreciate that Richard and others are willing to teach in this thread from the very beginning instead of trying to understand a finished circuit design.
I am not trying to pick on you, I just want to learn without all the negative responses to the cfa idea from the get go. I hope others who can help like Bob Cordell join in this thread as that is one of the reasons I liked his book so much, the way he presents the information from the simplest concepts to a final implementation.
Steven
Your statement about what you think this debate is about says it all. The debate was left in the other thread when we came to this new thread. This is not a debate thread, it is a thread to learn and understand the cfa circuit and how to build the best cfa that is possible, Leave the debate behind and either try and learn something, unless of course you don't need to and then why are you here? This isn't the bash the idea thread, please leave it at the door. If you have something useful to add to the cfa circuit great. It seems that you are impatient to get right to the final designs, but some of us are not anywhere near that level of understanding yet and I appreciate that Richard and others are willing to teach in this thread from the very beginning instead of trying to understand a finished circuit design.
I am not trying to pick on you, I just want to learn without all the negative responses to the cfa idea from the get go. I hope others who can help like Bob Cordell join in this thread as that is one of the reasons I liked his book so much, the way he presents the information from the simplest concepts to a final implementation.
Steven
Infinite, because the load is infinite. Any current source with an infinite load produces an infinite voltage.
But we should not confuse the practical implementation of an ideal source with its definition.
The definition of a 1A current source is a source that provides 1A whatever the load. Nothing more, nothing less.
jan
Last edited:
Yes it could be coined as voltage sag. But a perfect voltage source doesn't exist in real world. In real world, you cannot make the feedback so strong to avoid any voltage sag. You can also not come up with any capacitors to avoid voltage sag.
But this was not my point.
My point was -- oh well, just read the post
jan
You cannot improve the amp transient response with more power supply caps. The amplifier has it's own transient response, and as long as the power supply doesn't sag so much that the amp clips, it has almost no effect.
I say almost because, depending on the amp PSRR there may be a feedtrough to the output, but it is not impacting the amp transient response.
Square wave tests can be of interest for technical reasons but I agree they do not determine how the amp sounds.
The listening test is important but with listening test the problem is that it tells you whether you like the amp or not, NOT how good it reproduces the signal.
jan
And even here, we can have different approaches, from the musician who want the more natural sound as possible from his speakers, the sound engineer who want to reproduce all details from a tape, good and bad in the most transparent way, to can evaluate his work; the ones who listen music in a vertical way interested in sound stage, the ones in an horizontal one, following each instrument's melodic lines, interested in details reproduction and micro dynamic; the ones preferring horns and the ones condenser's enclosures...
And, here, no way to compare instant in and out to evaluate what an amplifier (as well as an enclosure) add or remove from an original source nobody know fully...
It need a long time experience, like learning wines or perfumes. Far from measurements numbers.